6 - Validacion de EcuacioneS KFRE Recalibradas en datos imputados

1 Code to reproduce results of the manuscript ‘Kidney Failure Prediction: Multicenter External Validation of the KFRE Model in Patients with CKD Stages 3-4 in Peru’

1.1 Introduction

This document presents the code and results of the main analysis shown in the article.

1.2 Setup

rm(list = ls())

# Use pacman to check whether packages are installed, if not load
if (!require("pacman")) install.packages("pacman")
library(pacman)

# Unload all package to begin in a session with only base packages
pacman::p_unload("all")

# Install packages
pacman::p_load(
  here, 
  skimr, 
  survival,
  survminer, 
  rms,
  cmprsk,
  riskRegression,
  mstate,
  pseudo,
  pec,
  plotrix,
  knitr,
  splines,
  kableExtra,
  flextable,
  gtsummary,
  boot,
  tidyverse,
  rsample,
  gridExtra,
  webshot, 
  patchwork,
  survival, 
  ggsci, 
  cowplot, 
  scales, 
  patchwork, 
  labelled, 
  glue, 
  dcurves, 
  broom, 
  downlit, 
  xml2, 
  gghalves, 
  devtools, 
  htmltools, 
  gghalves, 
  ggtext, 
  DiagrammeR, 
  gt, 
  janitor, 
  VIM, 
  PerformanceAnalytics, 
  mice, 
  rms, 
  naniar, 
  DescTools, 
  gtools, 
  ggExtra, 
  furrr, 
  future, 
  ggmice,
  parallel,
  tictoc, 
  rio,
  tidymodels
)

if (!require("impstep")) remotes::install_github("bgravesteijn/impstep", force = TRUE)
if (!require("smplot2")) devtools::install_github('smin95/smplot2', force = TRUE)

library(impstep)

1.2.0.1 Funciones

source(here("Code", "source", "kfre_pi.R"))
source(here("Code", "source", "kfre_pr.R"))
source(here("Code", "source", "oe_ratio.R"))
source(here("Code", "source", "calibration_intercept.R"))
source(here("Code", "source", "calibration_slope.R"))
source(here("Code", "source", "auc.R"))
source(here("Code", "source", "auc_brier_boot.R"))
source(here("Code", "source", "validate.mids.R"))
source(here("Code", "source", "pool.validate.mids.R"))
source(here("Code", "source", "pool.auc.mids.R"))
source(here("Code", "source", "process_imp_cal_plot.R"))
source(here("Code", "source", "predict.mira.R"))
source(here("Code", "source", "performance_measures.R"))
source(here("Code", "source", "tidy_performance_stack.R"))    
source(here("Code", "source", "tidy_pool.R")) 
source(here("Code", "source", "process_imp_cal_plot2.R")) 
source(here("Code", "source", "print_equation.R")) 

1.2.1 Importar datos

Las ecuaciones originales se muestran a continuacion:

eq_kfre_original2y <- data.frame(
  vars = c("age", "male", "eGFR", "logACR"),
  coefs = c(-0.2201, 0.2467, -0.5567, 0.4510),
  scale = c(10, 1, 5, 1),
  center = c(7.036, 0.5642, 7.222, 5.137)
)

st02y <- 0.9832


eq_kfre_original5y <- data.frame(
  vars = c("age", "male", "eGFR", "logACR"),
  coefs = c(-0.2201, 0.2467, -0.5567, 0.4510),
  scale = c(10, 1, 5, 1),
  center = c(7.036, 0.5642, 7.222, 5.137)
)
st05y <- 0.9365

• A 2 anios:

print_equation(eq_kfre_original2y, st02y)

\[1 - 0.9832^{exp(- 0.2201 \times (age / 10 - 7.036) + 0.2467 \times (male / 1 - 0.5642) - 0.5567 \times (eGFR / 5 - 7.222) + 0.451 \times (logACR / 1 - 5.137))}\]

• A 5 anios:

print_equation(eq_kfre_original2y, st02y)

\[1 - 0.9832^{exp(- 0.2201 \times (age / 10 - 7.036) + 0.2467 \times (male / 1 - 0.5642) - 0.5567 \times (eGFR / 5 - 7.222) + 0.451 \times (logACR / 1 - 5.137))}\]

Las nuevas sobrevidas basales y los factores de recalibracion de los coeficientes de las ecuaciones recalibradas, dependiendo del metodo usado, se muestran a continuacion

recal_loads <- import(here("Data", "Tidy", "equations", "recal_loads.rds")) 
recal_loads |> 
  kbl() |> 
  kable_classic()

metodo	year	sobrev_recal	fc_coef
A	2	0.9766676	1.0000000
A	5	0.9496968	1.0000000
B	2	0.9624290	0.7405622
B	5	0.9241750	0.7405622
C	2	0.9772796	1.0000000
C	5	0.9537293	1.0000000
D	2	0.9639901	0.7405622
D	5	0.9328086	0.7405622

1.3 Configurar cores para parallel processing

n_cores <- detectCores()
n_cores <- 3

# Evaluate futures in parallel - max of two workers to avoid hogging resources
future::plan("multisession", workers = n_cores)

1.4 Set some constants

seed <- 2014
primary_event <- 1
imputs <- 101

1.5 Importar datos

data_impA <- readRDS(here::here("Data", "Tidy", "data_impA.rds")) 

imp.datosA <- complete(data_impA, action = "long", include = TRUE) |> 
  filter(.imp != 0) |> 
  filter(.imp < imputs)

imp.datosA2 <- imp.datosA 

1.6 Metodo A: Reestimar riesgo basal usando Cox

source(here("Code", "source", "kfre_pi.R"))
source(here("Code", "source", "kfre_pr.R"))
source(here("Code", "source", "oe_ratio.R"))
source(here("Code", "source", "calibration_intercept.R"))
source(here("Code", "source", "calibration_slope.R"))
source(here("Code", "source", "auc.R"))
source(here("Code", "source", "auc_brier_boot.R"))
source(here("Code", "source", "validate.mids.R"))
source(here("Code", "source", "pool.validate.mids.R"))
source(here("Code", "source", "pool.auc.mids.R"))
source(here("Code", "source", "process_imp_cal_plot.R"))
source(here("Code", "source", "predict.mira.R"))
source(here("Code", "source", "performance_measures.R"))
source(here("Code", "source", "tidy_performance_stack.R"))    
source(here("Code", "source", "tidy_pool.R")) 
source(here("Code", "source", "process_imp_cal_plot2.R")) 
source(here("Code", "source", "print_equation.R")) 

df_recal_metA <- import(here("Data", "Tidy", "equations", "df_recal_modA.rds"))

df_recal_metA2y <- df_recal_metA |> 
  filter(year == 2) |> 
  select(-year) |> 
  rename(st0_imp2y = st0_imp, 
         fc_coef_imp2y = fc_coef_imp)

df_recal_metA5y <- df_recal_metA |> 
  filter(year == 5) |> 
  select(-year) |> 
  rename(st0_imp5y = st0_imp, 
         fc_coef_imp5y = fc_coef_imp)

rm(df_recal_metA)

imp.datosA <- imp.datosA2 |> 
  left_join(df_recal_metA2y, by = ".imp") |> 
  left_join(df_recal_metA5y, by = ".imp") |> 
  mutate(eventdf = factor(eventd), 
        risk2y = 1 - st0_imp2y ^ exp(fc_coef_imp2y * kfre_pi(imp.datosA2)), 
        risk5y = 1 - st0_imp5y ^ exp(fc_coef_imp5y * kfre_pi(imp.datosA2))) |> 
  select(.imp, .id, time, eventd, eventdf, risk2y, risk5y)

rm(df_recal_metA2y, df_recal_metA5y)

head(imp.datosA)

1.6.1 Calibration and Discrimination Measures

future::plan("multisession", workers = n_cores)
results_stack3a4_2y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 2, 
                                  primary_event = 1, 
                                  pred = "risk2y",
                                  seed = seed, 
                                  n_cores = n_cores)

gc()

            used  (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   3764208 201.1    7207968  385.0   7207968  385.0
Vcells 111756686 852.7  328868831 2509.1 328842219 2508.9

rio::export(results_stack3a4_2y , here("Data", "Tidy", 
                                       "res_valext_kfre_stack3a4_2y_metA.rds"))

future::plan("multisession", workers = n_cores)
results_stack3a4_5y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 5, 
                                  primary_event = 1, 
                                  pred = "risk5y",
                                  seed = seed, 
                                  n_cores = n_cores)

rio::export(results_stack3a4_5y, here("Data", "Tidy", 
                                      "res_valext_kfre_stack3a4_5y_metA.rds"))

gc()

            used  (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   3770181 201.4    7207968  385.0   7207968  385.0
Vcells 111774031 852.8  328868831 2509.1 328842219 2508.9

res_pool1 <- tidy_pool(results_stack3a4_2y) 

res_pool1 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0252035	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0273046	0.0254293	0.0291799	0.0000000	0.0000009	0.0000000	0.0000009	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.0800980	0.0099412	0.1502547	0.0252434	0.0012279	0.0000527	0.0012811	30030	19154.845	0.0433783	0.0415748	0.0001022	30031
OE difference	0.0021011	0.0001913	0.0040109	0.0310655	0.0000009	0.0000000	0.0000009	30030	21068.100	0.0371016	0.0357743	0.0000932	30031
Calibration Intercept	-0.4005583	-0.5323005	-0.2688161	0.0000000	0.0034786	0.0010230	0.0045118	30030	1745.445	0.2970238	0.2290041	0.0010129	30031
Calibration Slope	-0.3778972	-0.4304318	-0.3253627	0.0000000	0.0005392	0.0001764	0.0007173	30029	1498.875	0.3303551	0.2483210	0.0011663	30031
Logit AUC	2.2408323	2.1134404	2.3682242	0.0000000	0.0035868	0.0006288	0.0042219	30030	3735.045	0.1770471	0.1504163	0.0004948	30031
Brier	0.0239139	0.0223573	0.0254704	0.0000000	0.0000006	0.0000001	0.0000006	30030	8457.616	0.0987858	0.0899045	0.0002258	30031

res_pool2 <- tidy_pool(results_stack3a4_5y) 

res_pool2 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0453388	NaN	NaN	NA	NaN	0.0000002	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0476187	0.0450853	0.0501521	0.0000000	0.0000017	0.0000000	0.0000017	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.0490997	-0.0068232	0.1050225	0.0852733	0.0007368	0.0000763	0.0008139	30030	7847.022	0.1046463	0.0947329	0.0002427	30031
OE difference	0.0022799	-0.0003715	0.0049313	0.0919136	0.0000017	0.0000002	0.0000018	30030	8878.942	0.0950871	0.0868306	0.0002154	30031
Calibration Intercept	-0.3139026	-0.4158364	-0.2119688	0.0000000	0.0019600	0.0007323	0.0026996	30030	1243.766	0.3773411	0.2739634	0.0013844	30031
Calibration Slope	-0.3755688	-0.4164766	-0.3346610	0.0000000	0.0003190	0.0001147	0.0004348	30029	1311.106	0.3632975	0.2664844	0.0013192	30031
Logit AUC	2.0545824	1.9535722	2.1555925	0.0000000	0.0022533	0.0003970	0.0026543	30030	3706.552	0.1779701	0.1510820	0.0004984	30031
Brier	0.0383330	0.0362386	0.0404274	0.0000000	0.0000010	0.0000002	0.0000011	30030	4481.016	0.1563455	0.1352066	0.0004159	30031

tab_res_2y <- res_pool1 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "2 years")

tab_res_5y <- res_pool2 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "5 years")

tab_res0 <- tab_res_2y |>
  bind_rows(tab_res_5y)

tab_res0 |> 
  as_grouped_data(groups = "year") |> 
  as_grouped_data(groups = "grupo") |> 
  flextable::as_flextable(hide_grouplabel = TRUE) |> 
  autofit() |> 
  set_header_labels(
    year = "Time horizon", 
    term = "Performance measure", 
    measures = "Method A"
  ) |>  
  bold(j = 1) |> 
  set_caption("Table. Performance measures of KFRE in the external dataset of patients with CKD Stages 3a-4") |>  
  theme_booktabs() |>   
  bold(bold = TRUE, part = "header") -> table_perf_final

table_perf_final %>% 
  flextable::save_as_docx(path = here("Tables/Table_Imputed_Performance_metA.docx"))

table_perf_final

Time horizon	Performance measure	Method A
2 years
Calibration
	Average predicted risk	2.52%
	Overall observerd risk	2.73% (2.54% to 2.92%)
	OE ratio	1.08 (1.01 to 1.16)
	OE difference	0.21% (0.02% to 0.4%)
	Calibration Intercept	-0.4 (-0.53 to -0.27)
	Calibration Slope	-0.38 (-0.43 to -0.33)
Discrimination
	AUC	0.9 (0.89 to 0.91)
Overall performance
	Brier	0.02 (0.02 to 0.03)
5 years
Calibration
	Average predicted risk	4.53%
	Overall observerd risk	4.76% (4.51% to 5.02%)
	OE ratio	1.05 (0.99 to 1.11)
	OE difference	0.23% (-0.04% to 0.49%)
	Calibration Intercept	-0.31 (-0.42 to -0.21)
	Calibration Slope	-0.38 (-0.42 to -0.33)
Discrimination
	AUC	0.89 (0.88 to 0.9)
Overall performance
	Brier	0.04 (0.04 to 0.04)

rm(list=ls()[! ls() %in% c("imp.datosA", "imp.datosA2", "vdata", 
                           "primary_event", "horizon", 
                           "process_imp_cal_plot", "seed", "n_cores", "kfre_pi", 
                           "imputs")])
gc()

            used  (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   3929819 209.9    7207968  385.0   7207968  385.0
Vcells 101968334 778.0  328868831 2509.1 328842219 2508.9

1.6.2 Moderate calibration: Calibration curves lowess based on subdistributional hazards

primary_event <- 1

n_internal_knots <- 5

# Seleccion del grupo: Stages 3-4----

# A 2 años----
horizon <- 2

vdata <- imp.datosA %>% 
  rename(pred = risk2y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred)))

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_2y_metA.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_2y_metA.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue") + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_2y

# Grafico de calibracion con curvas imputadas
plot_mod0_2y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.1, color = "#38B8F7") -> plot_mod0_imp_2y

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4007492  214.1    7207968  385.0   7207968  385.0
Vcells 146986386 1121.5  328868831 2509.1 328842219 2508.9

# A 5 años----
horizon <- 5

vdata <- imp.datosA %>% 
  rename(pred = risk5y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred))) 

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_5y_metA.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_5y_metA.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue", 
             alpha = 0.5) + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_5y

# Grafico de calibracion con curvas imputadas
plot_mod0_5y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.3, color = "#38B8F7") -> plot_mod0_imp_5y

## Grafico final
plot_mod0_2y <- plot_mod0_2y + 
  labs(title = "Método A\n(2 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_5y <- plot_mod0_5y + 
  labs(title = "Método A\n(5 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_2y <- plot_mod0_imp_2y +
  labs(title = "Método A\n(2 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_5y <- plot_mod0_imp_5y +
  labs(title = "Método A\n(5 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_cal_mod0 <- plot_mod0_2y / plot_mod0_5y
plot_cal_imp_mod0 <- plot_mod0_imp_2y / plot_mod0_imp_5y


export(plot_mod0_2y, here("Data", "Tidy", "plot_metA_2y.rds"))
export(plot_mod0_5y, here("Data", "Tidy", "plot_metA_5y.rds"))
export(plot_cal_mod0, here("Data", "Tidy", "plot_cal_metA.rds"))

export(plot_mod0_imp_2y, here("Data", "Tidy", "plot_metA_imp_2y.rds"))
export(plot_mod0_imp_5y, here("Data", "Tidy", "plot_metA_imp_5y.rds"))
export(plot_cal_imp_mod0, here("Data", "Tidy", "plot_cal_imp_metA.rds"))


ggsave(filename = "Plot_Calibration_imputed_metA.png", 
       device = "png", 
       plot = plot_cal_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

ggsave(filename = "Plot_Calibration_imputed_estabilidad_metA.png", 
       device = "png", 
       plot = plot_cal_imp_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4083426  218.1    7207968  385.0   7207968  385.0
Vcells 159154711 1214.3  398349258 3039.2 398349258 3039.2

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_metA.png"))

Calibration curves for each group and prediction horizon

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_estabilidad_metA.png"))

Calibration curves for each group and prediction horizon

1.7 Metodo B: Reestimar coeficientes mediante Cox

source(here("Code", "source", "kfre_pi.R"))
source(here("Code", "source", "kfre_pr.R"))
source(here("Code", "source", "oe_ratio.R"))
source(here("Code", "source", "calibration_intercept.R"))
source(here("Code", "source", "calibration_slope.R"))
source(here("Code", "source", "auc.R"))
source(here("Code", "source", "auc_brier_boot.R"))
source(here("Code", "source", "validate.mids.R"))
source(here("Code", "source", "pool.validate.mids.R"))
source(here("Code", "source", "pool.auc.mids.R"))
source(here("Code", "source", "process_imp_cal_plot.R"))
source(here("Code", "source", "predict.mira.R"))
source(here("Code", "source", "performance_measures.R"))
source(here("Code", "source", "tidy_performance_stack.R"))    
source(here("Code", "source", "tidy_pool.R")) 
source(here("Code", "source", "process_imp_cal_plot2.R")) 
source(here("Code", "source", "print_equation.R")) 

df_recal_metB <- import(here("Data", "Tidy", "equations", "df_recal_modB.rds"))

df_recal_metB2y <- df_recal_metB |> 
  filter(year == 2) |> 
  select(-year) |> 
  rename(st0_imp2y = st0_imp, 
         fc_coef_imp2y = fc_coef_imp)

df_recal_metB5y <- df_recal_metB |> 
  filter(year == 5) |> 
  select(-year) |> 
  rename(st0_imp5y = st0_imp, 
         fc_coef_imp5y = fc_coef_imp)

rm(df_recal_metB)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087395  218.3    7207968  385.0   7207968  385.0
Vcells 159152528 1214.3  398349258 3039.2 398349258 3039.2

imp.datosA <- imp.datosA2 |> 
  left_join(df_recal_metB2y, by = ".imp") |> 
  left_join(df_recal_metB5y, by = ".imp") |> 
  mutate(eventdf = factor(eventd), 
        risk2y = 1 - st0_imp2y ^ exp(fc_coef_imp2y * kfre_pi(imp.datosA2)), 
        risk5y = 1 - st0_imp5y ^ exp(fc_coef_imp5y * kfre_pi(imp.datosA2))) |> 
  select(.imp, .id, time, eventd, eventdf, risk2y, risk5y)

rm(df_recal_metB2y, df_recal_metB5y)

head(imp.datosA)

1.7.1 Calibration and Discrimination Measures

future::plan("multisession", workers = n_cores)
results_stack3a4_2y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 2, 
                                  primary_event = 1, 
                                  pred = "risk2y",
                                  seed = seed, 
                                  n_cores = n_cores)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088101  218.4    7207968  385.0   7207968  385.0
Vcells 171170734 1306.0  398349258 3039.2 398349258 3039.2

rio::export(results_stack3a4_2y , here("Data", "Tidy", 
                                       "res_valext_kfre_stack3a4_2y_metB.rds"))

future::plan("multisession", workers = n_cores)
results_stack3a4_5y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 5, 
                                  primary_event = 1, 
                                  pred = "risk5y",
                                  seed = seed, 
                                  n_cores = n_cores)

rio::export(results_stack3a4_5y, here("Data", "Tidy", 
                                      "res_valext_kfre_stack3a4_5y_metB.rds"))

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087889  218.4    7207968  385.0   7207968  385.0
Vcells 171173949 1306.0  398349258 3039.2 398349258 3039.2

res_pool1 <- tidy_pool(results_stack3a4_2y) 

res_pool1 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0281081	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0273046	0.0254293	0.0291799	0.0000000	0.0000009	0.0000000	0.0000009	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	-0.0289994	-0.0978855	0.0398868	0.4093044	0.0012279	0.0000073	0.0012352	30030	29535.812	0.0059726	0.0059371	0.0000675	30031
OE difference	-0.0008035	-0.0026848	0.0010777	0.4025053	0.0000009	0.0000000	0.0000009	30030	29487.469	0.0063300	0.0062902	0.0000676	30031
Calibration Intercept	-0.0526479	-0.1503187	0.0450228	0.2907181	0.0023162	0.0001649	0.0024828	30030	12319.520	0.0719226	0.0670968	0.0001569	30031
Calibration Slope	-0.1599919	-0.2267812	-0.0932027	0.0000027	0.0009830	0.0001757	0.0011605	30029	3631.124	0.1804700	0.1528798	0.0005083	30031
Logit AUC	2.2408323	2.1134404	2.3682242	0.0000000	0.0035868	0.0006288	0.0042219	30030	3735.045	0.1770471	0.1504163	0.0004948	30031
Brier	0.0229702	0.0215023	0.0244380	0.0000000	0.0000005	0.0000000	0.0000006	30030	16502.247	0.0529381	0.0502765	0.0001181	30031

res_pool2 <- tidy_pool(results_stack3a4_5y) 

res_pool2 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0522702	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0476187	0.0450853	0.0501521	0.0000000	0.0000017	0.0000000	0.0000017	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	-0.0931965	-0.1467136	-0.0396795	0.0006427	0.0007368	0.0000087	0.0007455	30030	28499.275	0.0118740	0.0117347	0.0000698	30031
OE difference	-0.0046515	-0.0072030	-0.0021000	0.0003531	0.0000017	0.0000000	0.0000017	30030	27943.357	0.0143016	0.0140999	0.0000711	30031
Calibration Intercept	-0.1294301	-0.2049456	-0.0539146	0.0007836	0.0013454	0.0001372	0.0014840	30030	8009.119	0.1030272	0.0934040	0.0002380	30031
Calibration Slope	-0.1568453	-0.2073916	-0.1062989	0.0000000	0.0005815	0.0000824	0.0006648	30029	5089.568	0.1431575	0.1252299	0.0003681	30031
Logit AUC	2.0545824	1.9535722	2.1555925	0.0000000	0.0022533	0.0003970	0.0026543	30030	3706.552	0.1779701	0.1510820	0.0004984	30031
Brier	0.0370046	0.0351057	0.0389034	0.0000000	0.0000009	0.0000001	0.0000009	30030	8145.761	0.1016991	0.0923111	0.0002341	30031

tab_res_2y <- res_pool1 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "2 years")

tab_res_5y <- res_pool2 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "5 years")

tab_res0 <- tab_res_2y |>
  bind_rows(tab_res_5y)

tab_res0 |> 
  as_grouped_data(groups = "year") |> 
  as_grouped_data(groups = "grupo") |> 
  flextable::as_flextable(hide_grouplabel = TRUE) |> 
  autofit() |> 
  set_header_labels(
    year = "Time horizon", 
    term = "Performance measure", 
    measures = "Method B"
  ) |>  
  bold(j = 1) |> 
  set_caption("Table. Performance measures of KFRE in the external dataset of patients with CKD Stages 3a-4") |>  
  theme_booktabs() |>   
  bold(bold = TRUE, part = "header") -> table_perf_final

table_perf_final %>% 
  flextable::save_as_docx(path = here("Tables/Table_Imputed_Performance_metB.docx"))

table_perf_final

Time horizon	Performance measure	Method B
2 years
Calibration
	Average predicted risk	2.81%
	Overall observerd risk	2.73% (2.54% to 2.92%)
	OE ratio	0.97 (0.91 to 1.04)
	OE difference	-0.08% (-0.27% to 0.11%)
	Calibration Intercept	-0.05 (-0.15 to 0.05)
	Calibration Slope	-0.16 (-0.23 to -0.09)
Discrimination
	AUC	0.9 (0.89 to 0.91)
Overall performance
	Brier	0.02 (0.02 to 0.02)
5 years
Calibration
	Average predicted risk	5.23%
	Overall observerd risk	4.76% (4.51% to 5.02%)
	OE ratio	0.91 (0.86 to 0.96)
	OE difference	-0.47% (-0.72% to -0.21%)
	Calibration Intercept	-0.13 (-0.2 to -0.05)
	Calibration Slope	-0.16 (-0.21 to -0.11)
Discrimination
	AUC	0.89 (0.88 to 0.9)
Overall performance
	Brier	0.04 (0.04 to 0.04)

rm(list=ls()[! ls() %in% c("imp.datosA","imp.datosA2", "vdata", 
                           "primary_event", "horizon", 
                           "process_imp_cal_plot", "seed", "n_cores", "kfre_pi", 
                           "imputs")])
gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4082407  218.1    7207968  385.0   7207968  385.0
Vcells 140720942 1073.7  398349258 3039.2 398349258 3039.2

1.7.2 Moderate calibration: Calibration curves lowess based on subdistributional hazards

primary_event <- 1

n_internal_knots <- 5

# Seleccion del grupo: Stages 3-4----

# A 2 años----
horizon <- 2

vdata <- imp.datosA %>% 
  rename(pred = risk2y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred)))

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_2y_metB.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_2y_metB.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue") + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_2y

# Grafico de calibracion con curvas imputadas
plot_mod0_2y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.1, color = "#38B8F7") -> plot_mod0_imp_2y

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087878  218.4    7207968  385.0   7207968  385.0
Vcells 147163516 1122.8  398349258 3039.2 398349258 3039.2

# A 5 años----
horizon <- 5

vdata <- imp.datosA %>% 
  rename(pred = risk5y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred))) 

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_5y_metB.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_5y_metB.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue", 
             alpha = 0.5) + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_5y

# Grafico de calibracion con curvas imputadas
plot_mod0_5y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.3, color = "#38B8F7") -> plot_mod0_imp_5y

## Grafico final
plot_mod0_2y <- plot_mod0_2y + 
  labs(title = "Método B\n(2 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_5y <- plot_mod0_5y + 
  labs(title = "Método B\n(5 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_2y <- plot_mod0_imp_2y +
  labs(title = "Método B\n(2 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_5y <- plot_mod0_imp_5y +
  labs(title = "Método B\n(5 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_cal_mod0 <- plot_mod0_2y / plot_mod0_5y
plot_cal_imp_mod0 <- plot_mod0_imp_2y / plot_mod0_imp_5y


export(plot_mod0_2y, here("Data", "Tidy", "plot_metB_2y.rds"))
export(plot_mod0_5y, here("Data", "Tidy", "plot_metB_5y.rds"))
export(plot_cal_mod0, here("Data", "Tidy", "plot_cal_metB.rds"))

export(plot_mod0_imp_2y, here("Data", "Tidy", "plot_metB_imp_2y.rds"))
export(plot_mod0_imp_5y, here("Data", "Tidy", "plot_metB_imp_5y.rds"))
export(plot_cal_imp_mod0, here("Data", "Tidy", "plot_cal_imp_metB.rds"))


ggsave(filename = "Plot_Calibration_imputed_metB.png", 
       device = "png", 
       plot = plot_cal_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

ggsave(filename = "Plot_Calibration_imputed_estabilidad_metB.png", 
       device = "png", 
       plot = plot_cal_imp_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4086853  218.3    7207968  385.0   7207968  385.0
Vcells 159175274 1214.5  400182495 3053.2 400182495 3053.2

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_metB.png"))

Calibration curves for each group and prediction horizon

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_estabilidad_metB.png"))

Calibration curves for each group and prediction horizon

1.8 Metodo C: Reestimar riesgo basal usando Cause-specific Hazard Models

source(here("Code", "source", "kfre_pi.R"))
source(here("Code", "source", "kfre_pr.R"))
source(here("Code", "source", "oe_ratio.R"))
source(here("Code", "source", "calibration_intercept.R"))
source(here("Code", "source", "calibration_slope.R"))
source(here("Code", "source", "auc.R"))
source(here("Code", "source", "auc_brier_boot.R"))
source(here("Code", "source", "validate.mids.R"))
source(here("Code", "source", "pool.validate.mids.R"))
source(here("Code", "source", "pool.auc.mids.R"))
source(here("Code", "source", "process_imp_cal_plot.R"))
source(here("Code", "source", "predict.mira.R"))
source(here("Code", "source", "performance_measures.R"))
source(here("Code", "source", "tidy_performance_stack.R"))    
source(here("Code", "source", "tidy_pool.R")) 
source(here("Code", "source", "process_imp_cal_plot2.R")) 
source(here("Code", "source", "print_equation.R")) 

df_recal_metC <- import(here("Data", "Tidy", "equations", "df_recal_modC.rds"))

df_recal_metC2y <- df_recal_metC |> 
  filter(year == 2) |> 
  select(-year) |> 
  rename(st0_imp2y = st0_imp, 
         fc_coef_imp2y = fc_coef_imp)

df_recal_metC5y <- df_recal_metC |> 
  filter(year == 5) |> 
  select(-year) |> 
  rename(st0_imp5y = st0_imp, 
         fc_coef_imp5y = fc_coef_imp)

rm(df_recal_metC)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088096  218.4    7207968  385.0   7207968  385.0
Vcells 159164900 1214.4  400182495 3053.2 400182495 3053.2

imp.datosA <- imp.datosA2 |> 
  left_join(df_recal_metC2y, by = ".imp") |> 
  left_join(df_recal_metC5y, by = ".imp") |> 
  mutate(eventdf = factor(eventd), 
        risk2y = 1 - st0_imp2y ^ exp(fc_coef_imp2y * kfre_pi(imp.datosA2)), 
        risk5y = 1 - st0_imp5y ^ exp(fc_coef_imp5y * kfre_pi(imp.datosA2))) |> 
  select(.imp, .id, time, eventd, eventdf, risk2y, risk5y)

rm(df_recal_metC2y, df_recal_metC5y)

head(imp.datosA)

1.8.1 Calibration and Discrimination Measures

future::plan("multisession", workers = n_cores)
results_stack3a4_2y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 2, 
                                  primary_event = 1, 
                                  pred = "risk2y",
                                  seed = seed, 
                                  n_cores = n_cores)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088271  218.4    7207968  385.0   7207968  385.0
Vcells 171180740 1306.1  400182495 3053.2 400182495 3053.2

rio::export(results_stack3a4_2y , here("Data", "Tidy", 
                                       "res_valext_kfre_stack3a4_2y_metC.rds"))

future::plan("multisession", workers = n_cores)
results_stack3a4_5y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 5, 
                                  primary_event = 1, 
                                  pred = "risk5y",
                                  seed = seed, 
                                  n_cores = n_cores)

rio::export(results_stack3a4_5y, here("Data", "Tidy", 
                                      "res_valext_kfre_stack3a4_5y_metC.rds"))

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088345  218.4    7207968  385.0   7207968  385.0
Vcells 171184436 1306.1  400182495 3053.2 400182495 3053.2

res_pool1 <- tidy_pool(results_stack3a4_2y) 

res_pool1 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0246722	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0273046	0.0254293	0.0291799	0.0000000	0.0000009	0.0000000	0.0000009	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.1014030	0.0313286	0.1714775	0.0045673	0.0012279	0.0000498	0.0012781	30030	19885.008	0.0409386	0.0393285	0.0000986	30031
OE difference	0.0026324	0.0007259	0.0045389	0.0068087	0.0000009	0.0000000	0.0000009	30030	22189.845	0.0335540	0.0324646	0.0000887	30031
Calibration Intercept	-0.3736795	-0.5054974	-0.2418615	0.0000000	0.0034786	0.0010281	0.0045170	30030	1732.977	0.2985036	0.2298827	0.0010197	30031
Calibration Slope	-0.3778972	-0.4304318	-0.3253627	0.0000000	0.0005392	0.0001764	0.0007173	30029	1498.875	0.3303551	0.2483210	0.0011663	30031
Logit AUC	2.2408323	2.1134404	2.3682242	0.0000000	0.0035868	0.0006288	0.0042219	30030	3735.045	0.1770471	0.1504163	0.0004948	30031
Brier	0.0238678	0.0223098	0.0254258	0.0000000	0.0000006	0.0000001	0.0000006	30030	8623.332	0.0973001	0.0886722	0.0002216	30031

res_pool2 <- tidy_pool(results_stack3a4_5y) 

res_pool2 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0426431	NaN	NaN	NA	NaN	0.0000001	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0476187	0.0450853	0.0501521	0.0000000	0.0000017	0.0000000	0.0000017	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.1103923	0.0549497	0.1658349	0.0000956	0.0007368	0.0000626	0.0008000	30030	10073.240	0.0858295	0.0790451	0.0001906	30031
OE difference	0.0049757	0.0023561	0.0075952	0.0001975	0.0000017	0.0000001	0.0000018	30030	12870.036	0.0690142	0.0645587	0.0001503	30031
Calibration Intercept	-0.2282797	-0.3302904	-0.1262690	0.0000123	0.0019600	0.0007363	0.0027036	30030	1234.433	0.3794003	0.2750473	0.0013940	30031
Calibration Slope	-0.3755688	-0.4164766	-0.3346610	0.0000000	0.0003190	0.0001147	0.0004348	30029	1311.106	0.3632975	0.2664844	0.0013192	30031
Logit AUC	2.0545824	1.9535722	2.1555925	0.0000000	0.0022533	0.0003970	0.0026543	30030	3706.552	0.1779701	0.1510820	0.0004984	30031
Brier	0.0381672	0.0360744	0.0402600	0.0000000	0.0000010	0.0000001	0.0000011	30030	4815.814	0.1487632	0.1294986	0.0003882	30031

tab_res_2y <- res_pool1 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "2 years")

tab_res_5y <- res_pool2 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "5 years")

tab_res0 <- tab_res_2y |>
  bind_rows(tab_res_5y)

tab_res0 |> 
  as_grouped_data(groups = "year") |> 
  as_grouped_data(groups = "grupo") |> 
  flextable::as_flextable(hide_grouplabel = TRUE) |> 
  autofit() |> 
  set_header_labels(
    year = "Time horizon", 
    term = "Performance measure", 
    measures = "Method C"
  ) |>  
  bold(j = 1) |> 
  set_caption("Table. Performance measures of KFRE in the external dataset of patients with CKD Stages 3a-4") |>  
  theme_booktabs() |>   
  bold(bold = TRUE, part = "header") -> table_perf_final

table_perf_final %>% 
  flextable::save_as_docx(path = here("Tables/Table_Imputed_Performance_metC.docx"))

table_perf_final

Time horizon	Performance measure	Method C
2 years
Calibration
	Average predicted risk	2.47%
	Overall observerd risk	2.73% (2.54% to 2.92%)
	OE ratio	1.11 (1.03 to 1.19)
	OE difference	0.26% (0.07% to 0.45%)
	Calibration Intercept	-0.37 (-0.51 to -0.24)
	Calibration Slope	-0.38 (-0.43 to -0.33)
Discrimination
	AUC	0.9 (0.89 to 0.91)
Overall performance
	Brier	0.02 (0.02 to 0.03)
5 years
Calibration
	Average predicted risk	4.26%
	Overall observerd risk	4.76% (4.51% to 5.02%)
	OE ratio	1.12 (1.06 to 1.18)
	OE difference	0.5% (0.24% to 0.76%)
	Calibration Intercept	-0.23 (-0.33 to -0.13)
	Calibration Slope	-0.38 (-0.42 to -0.33)
Discrimination
	AUC	0.89 (0.88 to 0.9)
Overall performance
	Brier	0.04 (0.04 to 0.04)

rm(list=ls()[! ls() %in% c("imp.datosA", "imp.datosA2", "vdata", 
                           "primary_event", "horizon", 
                           "process_imp_cal_plot", "seed", "n_cores", "kfre_pi", 
                           "imputs")])
gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4082605  218.1    7207968  385.0   7207968  385.0
Vcells 140730982 1073.7  400182495 3053.2 400182495 3053.2

1.8.2 Moderate calibration: Calibration curves lowess based on subdistributional hazards

primary_event <- 1

n_internal_knots <- 5

# Seleccion del grupo: Stages 3-4----

# A 2 años----
horizon <- 2

vdata <- imp.datosA %>% 
  rename(pred = risk2y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred)))

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_2y_metC.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_2y_metC.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue") + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_2y

# Grafico de calibracion con curvas imputadas
plot_mod0_2y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.1, color = "#38B8F7") -> plot_mod0_imp_2y

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087943  218.4    7207968  385.0   7207968  385.0
Vcells 147173331 1122.9  400182495 3053.2 400182495 3053.2

# A 5 años----
horizon <- 5

vdata <- imp.datosA %>% 
  rename(pred = risk5y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred))) 

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_5y_metC.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_5y_metC.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue", 
             alpha = 0.5) + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_5y

# Grafico de calibracion con curvas imputadas
plot_mod0_5y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.3, color = "#38B8F7") -> plot_mod0_imp_5y

## Grafico final
plot_mod0_2y <- plot_mod0_2y + 
  labs(title = "Método C\n(2 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_5y <- plot_mod0_5y + 
  labs(title = "Método C\n(5 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_2y <- plot_mod0_imp_2y +
  labs(title = "Método C\n(2 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_5y <- plot_mod0_imp_5y +
  labs(title = "Método C\n(5 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_cal_mod0 <- plot_mod0_2y / plot_mod0_5y
plot_cal_imp_mod0 <- plot_mod0_imp_2y / plot_mod0_imp_5y


export(plot_mod0_2y, here("Data", "Tidy", "plot_metC_2y.rds"))
export(plot_mod0_5y, here("Data", "Tidy", "plot_metC_5y.rds"))
export(plot_cal_mod0, here("Data", "Tidy", "plot_cal_metC.rds"))

export(plot_mod0_imp_2y, here("Data", "Tidy", "plot_metC_imp_2y.rds"))
export(plot_mod0_imp_5y, here("Data", "Tidy", "plot_metC_imp_5y.rds"))
export(plot_cal_imp_mod0, here("Data", "Tidy", "plot_cal_imp_metC.rds"))


ggsave(filename = "Plot_Calibration_imputed_metC.png", 
       device = "png", 
       plot = plot_cal_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

ggsave(filename = "Plot_Calibration_imputed_estabilidad_metC.png", 
       device = "png", 
       plot = plot_cal_imp_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087052  218.3    7207968  385.0   7207968  385.0
Vcells 159185313 1214.5  401556731 3063.7 401556731 3063.7

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_metC.png"))

Calibration curves for each group and prediction horizon

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_estabilidad_metC.png"))

Calibration curves for each group and prediction horizon

1.9 Metodo D: Reestimar coeficientes mediante Cause-specific Hazard Models

source(here("Code", "source", "kfre_pi.R"))
source(here("Code", "source", "kfre_pr.R"))
source(here("Code", "source", "oe_ratio.R"))
source(here("Code", "source", "calibration_intercept.R"))
source(here("Code", "source", "calibration_slope.R"))
source(here("Code", "source", "auc.R"))
source(here("Code", "source", "auc_brier_boot.R"))
source(here("Code", "source", "validate.mids.R"))
source(here("Code", "source", "pool.validate.mids.R"))
source(here("Code", "source", "pool.auc.mids.R"))
source(here("Code", "source", "process_imp_cal_plot.R"))
source(here("Code", "source", "predict.mira.R"))
source(here("Code", "source", "performance_measures.R"))
source(here("Code", "source", "tidy_performance_stack.R"))    
source(here("Code", "source", "tidy_pool.R")) 
source(here("Code", "source", "process_imp_cal_plot2.R")) 
source(here("Code", "source", "print_equation.R")) 

df_recal_metD <- import(here("Data", "Tidy", "equations", "df_recal_modD.rds"))

df_recal_metD2y <- df_recal_metD |> 
  filter(year == 2) |> 
  select(-year) |> 
  rename(st0_imp2y = st0_imp, 
         fc_coef_imp2y = fc_coef_imp)

df_recal_metD5y <- df_recal_metD |> 
  filter(year == 5) |> 
  select(-year) |> 
  rename(st0_imp5y = st0_imp, 
         fc_coef_imp5y = fc_coef_imp)

rm(df_recal_metD)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088295  218.4    7207968  385.0   7207968  385.0
Vcells 159174941 1214.5  401556731 3063.7 401556731 3063.7

imp.datosA <- imp.datosA2 |> 
  left_join(df_recal_metD2y, by = ".imp") |> 
  left_join(df_recal_metD5y, by = ".imp") |> 
  mutate(eventdf = factor(eventd), 
        risk2y = 1 - st0_imp2y ^ exp(fc_coef_imp2y * kfre_pi(imp.datosA2)), 
        risk5y = 1 - st0_imp5y ^ exp(fc_coef_imp5y * kfre_pi(imp.datosA2))) |> 
  select(.imp, .id, time, eventd, eventdf, risk2y, risk5y)

rm(df_recal_metD2y, df_recal_metD5y)

head(imp.datosA)

1.9.1 Calibration and Discrimination Measures

future::plan("multisession", workers = n_cores)
results_stack3a4_2y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 2, 
                                  primary_event = 1, 
                                  pred = "risk2y",
                                  seed = seed, 
                                  n_cores = n_cores)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088240  218.4    7207968  385.0   7207968  385.0
Vcells 171190403 1306.1  401556731 3063.7 401556731 3063.7

rio::export(results_stack3a4_2y , here("Data", "Tidy", 
                                       "res_valext_kfre_stack3a4_2y_metD.rds"))

future::plan("multisession", workers = n_cores)
results_stack3a4_5y <- tidy_performance_stack(imp.datosA, 
                                  horizon = 5, 
                                  primary_event = 1, 
                                  pred = "risk5y",
                                  seed = seed, 
                                  n_cores = n_cores)

rio::export(results_stack3a4_5y, here("Data", "Tidy", 
                                      "res_valext_kfre_stack3a4_5y_metD.rds"))

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088293  218.4    7207968  385.0   7207968  385.0
Vcells 171194065 1306.2  401556731 3063.7 401556731 3063.7

res_pool1 <- tidy_pool(results_stack3a4_2y) 

res_pool1 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0270440	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0273046	0.0254293	0.0291799	0.0000000	0.0000009	0.0000000	0.0000009	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.0095915	-0.0592998	0.0784828	0.7849385	0.0012279	0.0000074	0.0012354	30030	29515.649	0.0061234	0.0060862	0.0000675	30031
OE difference	0.0002605	-0.0016204	0.0021415	0.7860096	0.0000009	0.0000000	0.0000009	30030	29530.865	0.0060098	0.0059739	0.0000675	30031
Calibration Intercept	-0.0094101	-0.1070162	0.0881960	0.8501139	0.0023162	0.0001617	0.0024796	30030	12583.365	0.0705086	0.0658646	0.0001537	30031
Calibration Slope	-0.1599914	-0.2267805	-0.0932022	0.0000027	0.0009830	0.0001756	0.0011604	30029	3631.310	0.1804638	0.1528753	0.0005082	30031
Logit AUC	2.2408323	2.1134404	2.3682242	0.0000000	0.0035868	0.0006288	0.0042219	30030	3735.045	0.1770471	0.1504163	0.0004948	30031
Brier	0.0229607	0.0214844	0.0244369	0.0000000	0.0000005	0.0000000	0.0000006	30030	16955.342	0.0512092	0.0487146	0.0001151	30031

res_pool2 <- tidy_pool(results_stack3a4_5y) 

res_pool2 |> 
  kbl() |> 
  kable_classic(full_width = T, html_font = "Cambria")

term	estimate	ll	ul	pval	ubar	b	t	dfcom	df	riv	lambda	fmi	n
Average predicted risk	0.0470960	NaN	NaN	NA	NaN	0.0000000	NaN	30030	NA	NaN	NaN	NaN	30031
Overall observerd risk	0.0476187	0.0450853	0.0501521	0.0000000	0.0000017	0.0000000	0.0000017	30030	30030.000	0.0000000	0.0000000	0.0000666	30031
Log OE ratio	0.0110421	-0.0424854	0.0645696	0.6859700	0.0007368	0.0000090	0.0007458	30030	28413.190	0.0122698	0.0121211	0.0000700	30031
OE difference	0.0005227	-0.0020259	0.0030713	0.6876847	0.0000017	0.0000000	0.0000017	30030	28472.085	0.0120000	0.0118577	0.0000698	30031
Calibration Intercept	-0.0039685	-0.0793289	0.0713919	0.9177852	0.0013454	0.0001312	0.0014780	30030	8487.042	0.0985189	0.0896834	0.0002250	30031
Calibration Slope	-0.1568453	-0.2073916	-0.1062989	0.0000000	0.0005815	0.0000824	0.0006648	30029	5089.568	0.1431575	0.1252299	0.0003681	30031
Logit AUC	2.0545824	1.9535722	2.1555925	0.0000000	0.0022533	0.0003970	0.0026543	30030	3706.552	0.1779701	0.1510820	0.0004984	30031
Brier	0.0369244	0.0350041	0.0388447	0.0000000	0.0000009	0.0000001	0.0000010	30030	9129.116	0.0930076	0.0850933	0.0002097	30031

tab_res_2y <- res_pool1 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "2 years")

tab_res_5y <- res_pool2 |> 
  select(term, estimate, ll, ul) |> 
  mutate(
    across(c(estimate, ll, ul), ~ if_else(term == "Log OE ratio", exp(.x), .x)), 
    across(c(estimate, ll, ul), ~ if_else(term == "Logit AUC", exp(.x) / (1  + exp(.x)), .x)), 
    term = if_else(term == "Log OE ratio", "OE ratio", term), 
    term = if_else(term == "Logit AUC", "AUC", term), 
    across(c(estimate, ll, ul), ~ if_else(term %in% 
                                            c("Average predicted risk", 
                                              "Overall observerd risk", 
                                              "OE difference"), 100 * .x, .x)), 
    across(c(estimate, ll, ul), ~ round(.x, 2)), 
    measures = case_when(term == "Average predicted risk" ~ str_glue("{estimate}%"), 
                       term %in% c("Overall observerd risk", "OE difference") ~ str_glue("{estimate}% ({ll}% to {ul}%)"),
                       TRUE ~ str_glue("{estimate} ({ll} to {ul})")
                       )
    ) |> 
  select(term, measures) |> 
  mutate(grupo = c(rep("Calibration", 6), "Discrimination", "Overall performance"), 
         year = "5 years")

tab_res0 <- tab_res_2y |>
  bind_rows(tab_res_5y)

tab_res0 |> 
  as_grouped_data(groups = "year") |> 
  as_grouped_data(groups = "grupo") |> 
  flextable::as_flextable(hide_grouplabel = TRUE) |> 
  autofit() |> 
  set_header_labels(
    year = "Time horizon", 
    term = "Performance measure", 
    measures = "Method D"
  ) |>  
  bold(j = 1) |> 
  set_caption("Table. Performance measures of KFRE in the external dataset of patients with CKD Stages 3a-4") |>  
  theme_booktabs() |>   
  bold(bold = TRUE, part = "header") -> table_perf_final

table_perf_final %>% 
  flextable::save_as_docx(path = here("Tables/Table_Imputed_Performance_metD.docx"))

table_perf_final

Time horizon	Performance measure	Method D
2 years
Calibration
	Average predicted risk	2.7%
	Overall observerd risk	2.73% (2.54% to 2.92%)
	OE ratio	1.01 (0.94 to 1.08)
	OE difference	0.03% (-0.16% to 0.21%)
	Calibration Intercept	-0.01 (-0.11 to 0.09)
	Calibration Slope	-0.16 (-0.23 to -0.09)
Discrimination
	AUC	0.9 (0.89 to 0.91)
Overall performance
	Brier	0.02 (0.02 to 0.02)
5 years
Calibration
	Average predicted risk	4.71%
	Overall observerd risk	4.76% (4.51% to 5.02%)
	OE ratio	1.01 (0.96 to 1.07)
	OE difference	0.05% (-0.2% to 0.31%)
	Calibration Intercept	0 (-0.08 to 0.07)
	Calibration Slope	-0.16 (-0.21 to -0.11)
Discrimination
	AUC	0.89 (0.88 to 0.9)
Overall performance
	Brier	0.04 (0.04 to 0.04)

rm(list=ls()[! ls() %in% c("imp.datosA", "imp.datosA2", "vdata", 
                           "primary_event", "horizon", 
                           "process_imp_cal_plot", "seed", "n_cores", "kfre_pi", 
                           "imputs")])
gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4082805  218.1    7207968  385.0   7207968  385.0
Vcells 140741039 1073.8  401556731 3063.7 401556731 3063.7

1.9.2 Moderate calibration: Calibration curves lowess based on subdistributional hazards

primary_event <- 1

n_internal_knots <- 5

# Seleccion del grupo: Stages 3-4----

# A 2 años----
horizon <- 2

vdata <- imp.datosA %>% 
  rename(pred = risk2y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred)))

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_2y_metD.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_2y_metD.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue") + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_2y

# Grafico de calibracion con curvas imputadas
plot_mod0_2y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.1, color = "#38B8F7") -> plot_mod0_imp_2y

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4088152  218.4    7207968  385.0   7207968  385.0
Vcells 147183407 1123.0  401556731 3063.7 401556731 3063.7

# A 5 años----
horizon <- 5

vdata <- imp.datosA %>% 
  rename(pred = risk5y) |> 
  select(.imp, .id, time, eventd, pred) |> 
  mutate(cll_pred = log(-log(1 - pred))) 

rcs_vdata <- ns(vdata$cll_pred, df = n_internal_knots + 1)
colnames(rcs_vdata) <- paste0("basisf_", colnames(rcs_vdata))
knots <- attr(rcs_vdata, "knots")
bound.knots <-  attr(rcs_vdata, "Boundary.knots")

pp <- seq(min(vdata$pred), max(vdata$pred), length.out = 1000)
cll_pp <- log(-log(1 - pp))
rcs_cll_pp <- ns(cll_pp, knots = knots, Boundary.knots = bound.knots)
colnames(rcs_cll_pp) <- paste0("basisf_", colnames(rcs_cll_pp))

vdata_bis_pp <- cbind(pp, as.data.frame(rcs_cll_pp))

future::plan("multisession", workers = n_cores)
subdist_df_imp <- future_map(1:max(vdata$.imp),
                         process_imp_cal_plot, 
                         vdata = vdata, 
                         primary_event = primary_event, 
                         horizon = horizon, 
                         type = "subdist_hazard", 
                         n_internal_knots = n_internal_knots, 
                         vdata_bis_pp, 
                         .options = furrr_options(seed = seed, 
                                                  packages = c("riskRegression", 
                                                               "survival", 
                                                               "splines", 
                                                               "cmprsk",
                                                               "tidyverse")), 
                         .progress = TRUE)

[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"
[1] "basisf_1" "basisf_2" "basisf_3" "basisf_4" "basisf_5" "basisf_6"

# 5 knots seems to give somewhat equivalent graph to pseudo method with bw = 0.05
subdist_df_imp_obs <- subdist_df_imp |> 
  bind_rows() |> 
  filter(type == "observed")

subdist_df_stack <- subdist_df_imp_obs |>
  group_by(.imp) |> 
  mutate(deciles_risk = as.integer(quantcut(risk, seq(0, 1, by = 0.1)))) |> 
  group_by(.imp, deciles_risk) |> 
  summarise(obs_mean_imp = mean(obs), 
            risk_mean_imp = mean(risk)) |> 
  group_by(deciles_risk) |> 
  summarise(obs_mean_pool = mean(obs_mean_imp), 
            risk_mean_pool = mean(risk_mean_imp))
  
subdist_df_imp_fix <- subdist_df_imp |> 
  bind_rows() |>  
  filter(type == "fixed") |> 
  arrange(risk) |> 
  summarise(obs_pool = mean(obs), 
            .by = risk) |> 
  mutate(deciles_risk = quantcut(risk, seq(0, 1, by = 0.1)))

rio::export(subdist_df_imp, here("Data", "Tidy", "subdist_df_imput_3a4_5y_metD.rds"))
rio::export(subdist_df_stack, here("Data", "Tidy", "subdist_df_deciles_3a4_5y_metD.rds"))

# Grafico de calibracion
ggplot() +
  geom_abline(intercept = 0, slope = 1, colour = "red", linetype = 2) + 
  geom_line(data = subdist_df_imp_fix, 
            aes(x = risk, y = obs_pool),
            alpha = 0.5, color = "black") +
  geom_point(data = subdist_df_stack,
             aes(x = risk_mean_pool, y = obs_mean_pool),
             shape = 23,
             stroke = 0.1,
             fill = "blue", 
             alpha = 0.5) + 
  scale_y_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  scale_x_continuous(breaks = seq(0, 1, 0.2), limits = c(0, 1)) +
  theme_bw() + 
  labs(x = "Predicted risks", y = "Observed outcome proportions") + 
  # geom_rug(data = subdist_df_stack,
  #          aes(x = risk_mean_pool, y = obs_mean_pool),
  #          alpha = 0.1) +
  coord_fixed(ratio = 1, expand = TRUE)  -> plot_mod0_5y

# Grafico de calibracion con curvas imputadas
plot_mod0_5y + 
  geom_line(data = subdist_df_imp_obs, 
            aes(x = risk, y = obs, group = .imp),
            alpha = 0.3, color = "#38B8F7") -> plot_mod0_imp_5y

## Grafico final
plot_mod0_2y <- plot_mod0_2y + 
  labs(title = "Método D\n(2 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_5y <- plot_mod0_5y + 
  labs(title = "Método D\n(5 year KFRE)") + 
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_2y <- plot_mod0_imp_2y +
  labs(title = "Método D\n(2 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_mod0_imp_5y <- plot_mod0_imp_5y +
  labs(title = "Método D\n(5 year KFRE)") +
  theme(plot.title = element_text(hjust = 0.5))

plot_cal_mod0 <- plot_mod0_2y / plot_mod0_5y
plot_cal_imp_mod0 <- plot_mod0_imp_2y / plot_mod0_imp_5y


export(plot_mod0_2y, here("Data", "Tidy", "plot_metD_2y.rds"))
export(plot_mod0_5y, here("Data", "Tidy", "plot_metD_5y.rds"))
export(plot_cal_mod0, here("Data", "Tidy", "plot_cal_metD.rds"))

export(plot_mod0_imp_2y, here("Data", "Tidy", "plot_metD_imp_2y.rds"))
export(plot_mod0_imp_5y, here("Data", "Tidy", "plot_metD_imp_5y.rds"))
export(plot_cal_imp_mod0, here("Data", "Tidy", "plot_cal_imp_metD.rds"))


ggsave(filename = "Plot_Calibration_imputed_metD.png", 
       device = "png", 
       plot = plot_cal_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

ggsave(filename = "Plot_Calibration_imputed_estabilidad_metD.png", 
       device = "png", 
       plot = plot_cal_imp_mod0, 
       path = here("Figures"), 
       scale = 2, 
       width = 12, 
       height = 12,
       units = "cm", 
       dpi = 600)

gc()

            used   (Mb) gc trigger   (Mb)  max used   (Mb)
Ncells   4087251  218.3    7207968  385.0   7207968  385.0
Vcells 159195375 1214.6  482095234 3678.1 481855685 3676.3

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_metD.png"))

Calibration curves for each group and prediction horizon

knitr::include_graphics(here("Figures", "Plot_Calibration_imputed_estabilidad_metD.png"))

Calibration curves for each group and prediction horizon

1.10 Comparacion de modelos

plot_cal_metA <- import(here("Data", "Tidy", "plot_cal_metA.rds"))
plot_cal_metB <- import(here("Data", "Tidy", "plot_cal_metB.rds"))
plot_cal_metC <- import(here("Data", "Tidy", "plot_cal_metC.rds"))
plot_cal_metD <- import(here("Data", "Tidy", "plot_cal_metD.rds"))

(plot_cal_metA | plot_cal_metB | plot_cal_metC | plot_cal_metD) + plot_annotation(tag_levels = "A") -> plot_comp_calib

ggsave(filename = "Plot_Compare_Methods.png", 
       plot = plot_comp_calib, 
       device = "png", 
       path = here("Figures"), 
       scale = 2.5, 
       width = 12, 
       height = 6, 
       units = "cm")

1.11 Ticket de Reprocubilidad

sessionInfo()

R version 4.3.3 (2024-02-29 ucrt)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 11 x64 (build 22631)

Matrix products: default


locale:
[1] LC_COLLATE=English_United States.utf8 
[2] LC_CTYPE=English_United States.utf8   
[3] LC_MONETARY=English_United States.utf8
[4] LC_NUMERIC=C                          
[5] LC_TIME=English_United States.utf8    

time zone: America/Lima
tzcode source: internal

attached base packages:
 [1] parallel  grid      splines   stats     graphics  grDevices utils    
 [8] datasets  methods   base     

other attached packages:
 [1] smplot2_0.1.0              impstep_0.1.0             
 [3] yardstick_1.3.1            workflowsets_1.1.0        
 [5] workflows_1.1.4            tune_1.2.0                
 [7] recipes_1.0.10             parsnip_1.2.1             
 [9] modeldata_1.3.0            infer_1.0.7               
[11] dials_1.2.1                tidymodels_1.2.0          
[13] rio_1.0.1                  tictoc_1.2.1              
[15] ggmice_0.1.0               furrr_0.3.1               
[17] future_1.33.2              ggExtra_0.10.1            
[19] gtools_3.9.5               DescTools_0.99.54         
[21] naniar_1.1.0               mice_3.16.0               
[23] PerformanceAnalytics_2.0.4 xts_0.13.2                
[25] zoo_1.8-12                 VIM_6.2.2                 
[27] colorspace_2.1-0           janitor_2.2.0             
[29] gt_0.10.1                  DiagrammeR_1.0.11         
[31] ggtext_0.1.2               htmltools_0.5.8           
[33] devtools_2.4.5             usethis_2.2.3             
[35] gghalves_0.1.4             xml2_1.3.6                
[37] downlit_0.4.3              broom_1.0.5               
[39] dcurves_0.4.0              glue_1.7.0                
[41] labelled_2.12.0            scales_1.3.0              
[43] cowplot_1.1.3              ggsci_3.0.3               
[45] patchwork_1.2.0            webshot_0.5.5             
[47] gridExtra_2.3              rsample_1.2.1             
[49] lubridate_1.9.3            forcats_1.0.0             
[51] stringr_1.5.1              dplyr_1.1.4               
[53] purrr_1.0.2                readr_2.1.5               
[55] tidyr_1.3.1                tibble_3.2.1              
[57] tidyverse_2.0.0            boot_1.3-30               
[59] gtsummary_1.7.2            flextable_0.9.5           
[61] kableExtra_1.4.0           knitr_1.45                
[63] plotrix_3.8-4              pec_2023.04.12            
[65] prodlim_2023.08.28         pseudo_1.4.3              
[67] geepack_1.3.10             KMsurv_0.1-5              
[69] mstate_0.3.2               riskRegression_2023.12.21 
[71] cmprsk_2.2-11              rms_6.8-0                 
[73] Hmisc_5.1-2                survminer_0.4.9           
[75] ggpubr_0.6.0               ggplot2_3.5.0             
[77] survival_3.5-8             skimr_2.1.5               
[79] here_1.0.1                

loaded via a namespace (and not attached):
  [1] R.methodsS3_1.8.2       gld_2.6.6               pacman_0.5.1           
  [4] urlchecker_1.0.1        nnet_7.3-19             TH.data_1.1-2          
  [7] vctrs_0.6.5             png_0.1-8               digest_0.6.35          
 [10] shape_1.4.6.1           proxy_0.4-27            Exact_3.2              
 [13] httpcode_0.3.0          parallelly_1.37.1       MASS_7.3-60.0.1        
 [16] fontLiberation_0.1.0    httpuv_1.6.15           foreach_1.5.2          
 [19] withr_3.0.0             xfun_0.43               ellipsis_0.3.2         
 [22] memoise_2.0.1           crul_1.4.0              MatrixModels_0.5-3     
 [25] profvis_0.3.8           systemfonts_1.0.6       ragg_1.3.0             
 [28] R.oo_1.26.0             DEoptimR_1.1-3          Formula_1.2-5          
 [31] promises_1.2.1          httr_1.4.7              rstatix_0.7.2          
 [34] globals_0.16.3          rstudioapi_0.16.0       pan_1.9                
 [37] miniUI_0.1.1.1          generics_0.1.3          base64enc_0.1-3        
 [40] curl_5.2.1              repr_1.1.7              quadprog_1.5-8         
 [43] xtable_1.8-4            evaluate_0.23           hms_1.1.3              
 [46] glmnet_4.1-8            visNetwork_2.1.2        readxl_1.4.3           
 [49] magrittr_2.0.3          lmtest_0.9-40           snakecase_0.11.1       
 [52] later_1.3.2             lattice_0.22-5          future.apply_1.11.2    
 [55] robustbase_0.99-2       SparseM_1.81            lhs_1.1.6              
 [58] class_7.3-22            pillar_1.9.0            nlme_3.1-164           
 [61] iterators_1.0.14        GPfit_1.0-8             compiler_4.3.3         
 [64] stringi_1.8.3           gower_1.0.1             jomo_2.7-6             
 [67] minqa_1.2.6             crayon_1.5.2            abind_1.4-5            
 [70] haven_2.5.4             sp_2.1-3                rootSolve_1.8.2.4      
 [73] sandwich_3.1-0          codetools_0.2-19        multcomp_1.4-25        
 [76] textshaping_0.3.7       openssl_2.1.1           e1071_1.7-14           
 [79] lmom_3.0                mime_0.12               Rcpp_1.0.12            
 [82] quantreg_5.97           DiceDesign_1.10         cellranger_1.1.0       
 [85] gridtext_0.1.5          utf8_1.2.4              lme4_1.1-35.2          
 [88] fs_1.6.3                listenv_0.9.1           checkmate_2.3.1        
 [91] pkgbuild_1.4.4          expm_0.999-9            ggsignif_0.6.4         
 [94] Matrix_1.6-5            tzdb_0.4.0              visdat_0.6.0           
 [97] svglite_2.1.3           pkgconfig_2.0.3         tools_4.3.3            
[100] cachem_1.0.8            viridisLite_0.4.2       numDeriv_2016.8-1.1    
[103] fastmap_1.1.1           rmarkdown_2.26          sdamr_0.2.0            
[106] mets_1.3.4              officer_0.6.5           carData_3.0-5          
[109] rpart_4.1.23            farver_2.1.1            yaml_2.3.8             
[112] foreign_0.8-86          cli_3.6.2               lifecycle_1.0.4        
[115] askpass_1.2.0           mvtnorm_1.2-4           lava_1.8.0             
[118] sessioninfo_1.2.2       backports_1.4.1         timechange_0.3.0       
[121] gtable_0.3.4            jsonlite_1.8.8          mitml_0.4-5            
[124] pwr_1.3-0               zip_2.3.1               ranger_0.16.0          
[127] highr_0.10              polspline_1.1.24        survMisc_0.5.6         
[130] R.utils_2.12.3          timeDate_4032.109       shiny_1.8.1            
[133] gfonts_0.2.0            timereg_2.0.5           broom.helpers_1.14.0   
[136] gdtools_0.3.7           rprojroot_2.0.4         R6_2.5.1               
[139] km.ci_0.5-6             vcd_1.4-12              cluster_2.1.6          
[142] pkgload_1.3.4           ipred_0.9-14            nloptr_2.0.3           
[145] tidyselect_1.2.1        htmlTable_2.4.2         fontBitstreamVera_0.1.1
[148] car_3.1-2               munsell_0.5.0           laeken_0.5.3           
[151] fontquiver_0.2.1        data.table_1.15.4       htmlwidgets_1.6.4      
[154] RColorBrewer_1.1-3      rlang_1.1.3             uuid_1.2-0             
[157] remotes_2.5.0           fansi_1.0.6             hardhat_1.3.1